Due to WFLY-3359 you can run out of heap space on slave HCs when installing large patches from the DC. A patch slightly over 128mb will temporarily need 384mb of space as the backing byte array is resized, likely causing an OOME with the default -Xmx512m setting.
Maybe this should be ported back to 6.2.x though? I don't recall if that happened.
It's not in 6.2.x currently
verified on EAP 6.3.0.ER8 but not during ER8 test cycle. BZ was moved to ON_QA after ER8 test cycle ended.